Trademark law with Artificial Intelligence: An Alert or A Safeguard

Introduction

One of the key objectives of law is to reduce vagueness and promote clarity between subject matters. Trademark law is one of the best examples of achieving this objective. The primary objective of Trademark law is to indicate the commercial source of a product or service and in consequence that distinguishes one source from another. This differentiation is important because humans have inherent ‘faults’ that are failing to enable them to distinguish one source from another. Trademark law here helps to mitigate this gap of ‘faults’. However, with the advancement in the field of technology, this gap of inherent ‘faults’ is slowly being filled and the objective of trademark law is slowly being faded away. The inception of Artificial Intelligence (AI) has revolutionized the technical domain and it is impacting every field of human interactions and intellectual property laws are no exceptions. AI is gradually replacing human activities by providing ease of doing things efficiently. Thus, in the commercial domain, AI is becoming a ‘Middle Man’ between the producers or providers and consumers. Therefore, an obvious question arises, whether an AI has any inherent ‘faults’ like humans that may require to be addressed by Trademark law. This article will discuss this issue on two fronts—

1. Whether an AI can be registered as a trademark

2. What are the impacts of AI on the fundamental and well-accepted concepts of trademark law?

Whether an AI can be registered as a Trademark

A trademark can be registered in respect of any ‘Goods’ or ‘Services’. As per definitions under Section 2(1)(j) of the Trademarks Act, 1999, any subject matter of trade or manufacture is considered to be ‘Goods’. Now it is a well-settled and accepted principle that a person cannot be considered as a ‘Good’ and become a subject matter of trademark as it is against public morality. The Patents Act, 1970 in this respect, completely and expressly bars such consideration [Section 3(b) of the Patents Act, 1970]. If that is the case, then one may very well argue the validity of AI being registered as a trademark because there are ample examples where AI has been given the status of a human being such as Saudi Arabia granted citizenship to a humanoid robot¹ or an AI character was made an official resident of a busy central Tokyo district^2, etc. Therefore this matter is debatable and it may vary from jurisdiction to jurisdiction. But, if, for the sake of argument, AI cannot be considered as a good, can it be considered as a subject matter of ‘Services’? Here the answer might be negative as AI is assisting the beneficiary in achieving his desirable outcome through a process; the AI itself is not the service. Therefore, the trademark-ability of an AI is a disputed issue. However, in this regard, one must take note of the fact that, as per NICE classification of trademarks, humanoid robots with artificial intelligence is classified as ‘Goods’ under class 9.

Impacts of AI on the fundamental and well-accepted concepts of Trademark Law

Trademark law grants negative rights to the proprietor of the mark. When these rights are violated by any person other than the proprietor or the authorized persons, a cause of action arises and the right to remedy accrues on the proprietor. These remedies are two types— remedy against Infringement and remedy against Passing off. In both these cases, some fundamental and core concepts are discovered and developed through judicial pronouncements. These core concepts include ‘likelihood of confusion’, ‘average consumer’, ‘imperfect recollection’, ‘aural, conceptual and visual similarities’ etc. These are the inherent ‘faults’ in a human being that must be taken into considerations while deciding a case of infringement or passing off. Trademark law has withstood the three revolutions self-service, E-commerce, and social media but the question that remained unanswered is can it deal with the AI giant?³ It is already mentioned above, as AI is gradually replacing a common average consumer, are these core concepts of trademark law immune from the impact of AI? The validity of this argument comes from the fact that in this era of the World Wide Web (WWW) every single information is available to an AI to be scrutinized. Thus theoretically, there is ideally no chance of ‘likelihood of confusion’ for an AI, and every recollection by it is a perfect recollection.

However, there are certain serious ‘faults’ that have surfaced on many practical occasions. Firstly, it was discovered that AI can be confused. AI in several situations failed to distinguish aural, conceptual, and visual similarities. For example, Indian Trademark law allows word-marks where any word, even names, can be registered. Suppose two organizations are having registered word-marks ‘SONY’ and ‘SONI’. ‘SONY’ is a registered trademark under the Goods category and ‘SONI’ is a registered trademark under the Services category. Now, these two words are phonetically identical. Can AI be able to distinguish between these two trademarks if no visual input is allowed?

Secondly, the issue of external interference is hampering the process of independent autonomous decision making by an AI system. For example, when a consumer expresses his interest to buy certain products or to avail of certain services from an online shopping platform, the AI system built into that shopping website suggests relevant products or services. An obvious question arises, on what basis the AI is suggesting such products or services? Sometimes it is also seen that only products or services from certain brands are recommended again and again. What if only those brands are externally inserted into the AI system creating an ‘inherent bias’ into the system? If it is so, then an important question arises— what if AI recommends or suggests and purchases the counterfeits; can it be termed as a secondary infringer?⁴

Thirdly, an AI system is a predictive system rather than a reactive system.⁵ It predicts suggestions based on analysis of the previous search data of the consumer, price range, delivery location, and availability. It is very unlikely for an AI to understand the brand value, quality, intensity, and emotion of the consumer-related to a product or a service. For example, a consumer in India may be looking for a particular model of ‘NOKIA’ branded phone. There may be a chance that the consumer is a loyal consumer to that brand. Instead, the AI is recommending some cheap phones with better specifications. The question arises, can an AI possibly think like a human and recognize the brand value? If not, then how will it be able to restrain itself from violating the rights of a well-known trademark by way of passing off?

To date, the AI has not been able to successfully address these issues affirmatively. Therefore, it is still highly unlikely that brands or trademarks are free from infringement by AI.

Conclusion

It seems from the above discussions and elaborated examples that an AI system is still not capable enough to replace the jurisprudential basis of the Trademark law. The law has developed over centuries of trial and error in implementation and survived this long through numerous adaptations and modifications. Whereas, the concept of Artificial Intelligence is still in its nascent stage. But even if it catches up, it is very difficult for AI to completely shake off the foundation of a law which possess such concrete fundamentals. After all, an AI can be exceptionally efficient than a human being, but it certainly cannot possess an intelligence level higher than that of a human being. However, AI can assist in the domain of Trademark law which may furnish some exceptional output in the efficacy of the overall domain. Therefore the ‘Mantra’ should be— let not Artificial Intelligence be an obstacle to the Trademark law; rather let it be a supplement to the law.

BY-

DEBKRIPA BURMAN